--- Log opened Mon Mar 10 00:00:05 2014 | ||
--- Day changed Mon Mar 10 2014 | ||
@HeikoS | lambday: in the b-test paper code? | 00:00 |
---|---|---|
lambday | HeikoS: yeah... oh and they used 0 mean as well | 00:00 |
@HeikoS | lambday: so thats exactly what we do for the linear time mmd | 00:00 |
@HeikoS | lambday: just that this is not really a linear time estimate ;) | 00:00 |
@HeikoS | lambday: they cheated there | 00:01 |
@HeikoS | we would just use the diagonal of this matrix | 00:01 |
@HeikoS | lambday: or in fact block averages | 00:01 |
@HeikoS | lambday: can you send me the githuvb page? | 00:02 |
lambday | HeikoS: yeah.. just a sec | 00:02 |
lambday | HeikoS: https://github.com/wojzaremba/btest | 00:04 |
@HeikoS | lambday: yeah, so they cheat for the variance | 00:04 |
lambday | HeikoS: http://en.wikipedia.org/wiki/Algorithms_for_calculating_variance Parallel algorithm - | 00:04 |
lambday | talks about blocks | 00:04 |
@HeikoS | lambday: oh yeah, its totally possible to compute the blocks using multicore, everything is independent | 00:05 |
@HeikoS | lambday: good idea and extension :) | 00:05 |
@HeikoS | but check the btest.m file first | 00:05 |
lambday | HeikoS: yeah | 00:05 |
lambday | so that implementation is incorrent? | 00:05 |
@HeikoS | there they iterate over the blocks | 00:05 |
@HeikoS | lambday: nono its fine | 00:05 |
@HeikoS | but they compute a h-term for every block | 00:06 |
lambday | yeah | 00:06 |
@HeikoS | so that mean(hh) is the simple average of the blocks | 00:06 |
@HeikoS | and var is the covariance (computed in quadratic time) | 00:06 |
@HeikoS | we compute the variance in linear time via doing running averages, thats very similar to what I did in the linear time MMD code | 00:07 |
@HeikoS | for computing the covariance matrix between the kernels | 00:07 |
lambday | HeikoS: umm I am confused - do you mean https://github.com/lambday/shogun/blob/develop/src/shogun/statistics/LinearTimeMMD.cpp#L226 this is how it should be done? | 00:16 |
@HeikoS | lambday: yeah, this is just a running average of the individual statistic values | 00:17 |
lambday | HeikoS: this one for computing covariance matrix - https://github.com/lambday/shogun/blob/develop/src/shogun/statistics/LinearTimeMMD.cpp#L505 | 00:17 |
@HeikoS | lambday: for covariance matrix its also similar, since this covariance is between kernels | 00:17 |
lambday | yeah! | 00:17 |
@HeikoS | I think the code should be re-usable, just compouting the statistic should be different | 00:17 |
@HeikoS | first it was running average/variance of these linear time estimates | 00:18 |
@HeikoS | now it is a running average/variance of those quadratic time block estimates | 00:18 |
@HeikoS | lambday: but it will be very easy to verify - using the permutation | 00:18 |
lambday | HeikoS: ah this part has to be copied - I can't think of any clever way to separate this nicely :-/ | 00:21 |
lambday | HeikoS: and I didn't get the verify point :( | 00:21 |
@HeikoS | lambday: why copy? | 00:22 |
@HeikoS | lambday: should be done in streaming mmd I guess | 00:22 |
@HeikoS | that was the whole point of having a shared base class wasnt it? | 00:22 |
lambday | HeikoS: put that part into another function? | 00:22 |
@HeikoS | lambday: yeah | 00:22 |
@HeikoS | both linear time mmd and btest just do this online mean/variance of something | 00:23 |
lambday | HeikoS: alright - shall we do the same for covariance as well? cause we'd be computing that between kernels for btest too, right? | 00:24 |
lambday | HeikoS: I mean, move that online algorithm into another function | 00:24 |
@HeikoS | lambday: yeah! | 00:25 |
@HeikoS | lambday: and for the verification: | 00:25 |
@HeikoS | lambday: we can just compute the null distribution via the permutation test | 00:25 |
@HeikoS | lambday: which gives the ground truth | 00:25 |
@HeikoS | lambday: and then we can compare this to the Gaussian from variance approximation | 00:26 |
lambday | HeikoS: oh got it | 00:26 |
-!- gif [~gif@130.195.253.57] has joined #shogun | 00:27 | |
lambday | HeikoS: well I think both these methods can be moved to parent - things would be exactly same (except for the value of m2 which would be m/2 for linear and m for btest, which we can just pass via an additional arg)... things are exactly same! | 00:42 |
@HeikoS | lambday: yeah, indeed | 00:43 |
@HeikoS | lambday: I have to go to bed, see you tomorrow! | 00:45 |
lambday | HeikoS: okay :) good night man! | 00:45 |
@HeikoS | lambday: good night | 00:46 |
-!- HeikoS [~heiko@ip-176-199-255-21.unitymediagroup.de] has quit [Quit: Leaving.] | 00:46 | |
-!- lambday [67157f54@gateway/web/freenode/ip.103.21.127.84] has quit [] | 00:47 | |
-!- gif [~gif@130.195.253.57] has quit [Ping timeout: 264 seconds] | 01:08 | |
-!- abinash_panda [75ef5e6f@gateway/web/freenode/ip.117.239.94.111] has joined #shogun | 01:27 | |
-!- gif [~gif@130.195.253.57] has joined #shogun | 01:57 | |
-!- gif_ [~gif@130.195.253.57] has joined #shogun | 02:02 | |
-!- gif [~gif@130.195.253.57] has quit [Read error: Connection reset by peer] | 02:02 | |
-!- gif_ [~gif@130.195.253.57] has quit [Remote host closed the connection] | 02:32 | |
-!- gif [~gif@130.195.253.57] has joined #shogun | 02:33 | |
-!- sunil1337 [0e8bd057@gateway/web/freenode/ip.14.139.208.87] has joined #shogun | 02:44 | |
-!- sunil1337 [0e8bd057@gateway/web/freenode/ip.14.139.208.87] has left #shogun [] | 02:46 | |
-!- gif [~gif@130.195.253.57] has quit [Remote host closed the connection] | 03:06 | |
-!- gif [~gif@130.195.253.57] has joined #shogun | 04:03 | |
-!- gif [~gif@130.195.253.57] has quit [Remote host closed the connection] | 04:05 | |
-!- gif [~gif@130.195.253.57] has joined #shogun | 04:09 | |
-!- gif [~gif@130.195.253.57] has quit [Remote host closed the connection] | 04:10 | |
-!- pranet [73f91219@gateway/web/freenode/ip.115.249.18.25] has joined #shogun | 04:22 | |
-!- luckyjoon [12bd77d6@gateway/web/freenode/ip.18.189.119.214] has joined #shogun | 05:46 | |
luckyjoon | hello | 05:47 |
-!- pranet [73f91219@gateway/web/freenode/ip.115.249.18.25] has quit [Quit: Page closed] | 05:51 | |
-!- luckyjoon [12bd77d6@gateway/web/freenode/ip.18.189.119.214] has quit [Quit: Page closed] | 06:00 | |
-!- witness___ [uid10044@gateway/web/irccloud.com/x-znsfgwuamstauosi] has joined #shogun | 06:14 | |
witness___ | hi | 06:14 |
-!- tnosovich [~tnosovich@91.217.162.238] has quit [Ping timeout: 264 seconds] | 07:10 | |
-!- abinash_panda [75ef5e6f@gateway/web/freenode/ip.117.239.94.111] has quit [Ping timeout: 245 seconds] | 07:59 | |
-!- sonne|osx [~sonne@89.204.139.157] has joined #shogun | 08:21 | |
-!- sonne|osx [~sonne@89.204.139.157] has quit [Quit: sonne|osx] | 08:37 | |
-!- abinash_panda [0e8b2997@gateway/web/freenode/ip.14.139.41.151] has joined #shogun | 08:55 | |
@besser82 | wiking, lisitsyn: is there any specific reason to have the makefile in examples-dir to be purged? | 09:33 |
-!- kislay [~kislay@93.114.45.13] has joined #shogun | 10:20 | |
@wiking | besser82: yeah i thought there's no need for it... and if so we can reproduce that by a cmakefile or? | 10:56 |
-!- kprah [67157d4c@gateway/web/freenode/ip.103.21.125.76] has joined #shogun | 11:01 | |
kprah | were we supposed to have a meet today? | 11:02 |
@wiking | see the dopic | 11:02 |
@wiking | *topic | 11:02 |
@wiking | there's every monday a general session | 11:02 |
kprah | ok. | 11:03 |
kprah | Today's topic is GSOC/slammtisch then? | 11:04 |
-!- kislay [~kislay@93.114.45.13] has quit [Ping timeout: 264 seconds] | 11:06 | |
-!- kprah [67157d4c@gateway/web/freenode/ip.103.21.125.76] has quit [Ping timeout: 245 seconds] | 11:08 | |
-!- abinash_panda [0e8b2997@gateway/web/freenode/ip.14.139.41.151] has quit [Ping timeout: 245 seconds] | 11:15 | |
-!- kislay [~kislay@93.114.45.13] has joined #shogun | 11:51 | |
-!- PirosB3 [~pirosb3@162.219.6.191] has joined #shogun | 11:53 | |
PirosB3 | goodmorning | 11:53 |
PirosB3 | hi sonney2k thanks for the comments on the PR. why should we not use init_hogun exit_shogun in modular? | 11:53 |
-!- kislay [~kislay@93.114.45.13] has quit [Ping timeout: 264 seconds] | 11:56 | |
-!- lisitsyn [~lisitsyn@109-226-71-150.clients.tlt.100megabit.ru] has quit [Quit: Leaving.] | 12:33 | |
-!- lisitsyn [~lisitsyn@109-226-71-150.clients.tlt.100megabit.ru] has joined #shogun | 12:33 | |
-!- ashwink [67157e57@gateway/web/freenode/ip.103.21.126.87] has joined #shogun | 13:50 | |
-!- ashwink [67157e57@gateway/web/freenode/ip.103.21.126.87] has left #shogun [] | 13:51 | |
-!- ashwink [67157e57@gateway/web/freenode/ip.103.21.126.87] has joined #shogun | 13:51 | |
-!- ashwink [67157e57@gateway/web/freenode/ip.103.21.126.87] has left #shogun [] | 13:51 | |
-!- ashwink [67157e57@gateway/web/freenode/ip.103.21.126.87] has joined #shogun | 14:04 | |
-!- ashwink [67157e57@gateway/web/freenode/ip.103.21.126.87] has left #shogun [] | 14:05 | |
-!- PirosB3 [~pirosb3@162.219.6.191] has quit [Quit: PirosB3] | 14:06 | |
-!- dhruv13 [~dhruv13@84.64.13.90] has joined #shogun | 14:09 | |
-!- sonne|work [~sonnenbu@24-134-74-216-dynip.superkabel.de] has joined #shogun | 14:24 | |
-!- kislay [~kislay@93.114.45.13] has joined #shogun | 14:29 | |
-!- PirosB3 [~pirosb3@162.219.6.191] has joined #shogun | 14:44 | |
-!- PirosB3 [~pirosb3@162.219.6.191] has quit [Quit: PirosB3] | 14:57 | |
-!- PirosB3 [~pirosb3@162.219.6.191] has joined #shogun | 15:06 | |
-!- Sunil_ [0e8bd057@gateway/web/freenode/ip.14.139.208.87] has joined #shogun | 15:20 | |
-!- PirosB3 [~pirosb3@162.219.6.191] has quit [Quit: PirosB3] | 15:21 | |
-!- dhruv13 [~dhruv13@84.64.13.90] has quit [Ping timeout: 240 seconds] | 15:23 | |
-!- Sunil_ [0e8bd057@gateway/web/freenode/ip.14.139.208.87] has quit [Ping timeout: 245 seconds] | 15:24 | |
-!- PirosB3 [~pirosb3@162.219.6.191] has joined #shogun | 15:26 | |
-!- PirosB3 [~pirosb3@162.219.6.191] has quit [Client Quit] | 15:27 | |
-!- abinash_panda [0126170c@gateway/web/freenode/ip.1.38.23.12] has joined #shogun | 15:53 | |
-!- FSCV [~FSCV@189.203.250.138] has joined #shogun | 16:38 | |
-!- FSCV [~FSCV@189.203.250.138] has quit [Client Quit] | 16:39 | |
-!- jiaolong [9e6d09e4@gateway/web/freenode/ip.158.109.9.228] has joined #shogun | 16:49 | |
-!- jiaolong [9e6d09e4@gateway/web/freenode/ip.158.109.9.228] has left #shogun [] | 17:00 | |
-!- jiaolong [9e6d09e4@gateway/web/freenode/ip.158.109.9.228] has joined #shogun | 17:01 | |
-!- witness___ [uid10044@gateway/web/irccloud.com/x-znsfgwuamstauosi] has quit [Quit: Connection closed for inactivity] | 17:04 | |
-!- [paladin] [73f91219@gateway/web/freenode/ip.115.249.18.25] has joined #shogun | 17:50 | |
-!- kprah [67157d4c@gateway/web/freenode/ip.103.21.125.76] has joined #shogun | 18:14 | |
-!- kprah [67157d4c@gateway/web/freenode/ip.103.21.125.76] has quit [Ping timeout: 245 seconds] | 18:22 | |
-!- kprah [67157f4c@gateway/web/freenode/ip.103.21.127.76] has joined #shogun | 18:35 | |
-!- kprah [67157f4c@gateway/web/freenode/ip.103.21.127.76] has quit [Client Quit] | 18:36 | |
-!- pickle27 [d8bfe7da@gateway/web/freenode/ip.216.191.231.218] has joined #shogun | 18:46 | |
-!- PirosB3 [~pirosb3@162.219.6.191] has joined #shogun | 18:52 | |
-!- zxtx [~zv@c-98-223-196-32.hsd1.in.comcast.net] has quit [Ping timeout: 240 seconds] | 19:08 | |
-!- zxtx [~zv@c-98-223-196-32.hsd1.in.comcast.net] has joined #shogun | 19:23 | |
-!- witness___ [uid10044@gateway/web/irccloud.com/x-ylzuqzgmsbzbgkfd] has joined #shogun | 19:28 | |
-!- bumblebee [~quassel@202.78.172.162] has joined #shogun | 19:37 | |
-!- bumblebee is now known as Guest31547 | 19:37 | |
-!- Guest31547 [~quassel@202.78.172.162] has quit [Client Quit] | 19:37 | |
-!- bumblebee_ [~quassel@202.78.172.162] has joined #shogun | 19:38 | |
bumblebee_ | just wanted to know, while making a pull request, i send it from "username/my_personal_develop" branch to the main "shogun/develop" branch? | 19:45 |
-!- Cheparukhin [b064f6fe@gateway/web/freenode/ip.176.100.246.254] has joined #shogun | 19:46 | |
PirosB3 | hey guys, are we doing Stammtisch today? | 19:50 |
-!- HeikoS [~heiko@p4FED46C2.dip0.t-ipconnect.de] has joined #shogun | 19:51 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 19:51 | |
-!- bumblebee_ [~quassel@202.78.172.162] has quit [Ping timeout: 252 seconds] | 20:01 | |
-!- dhruv13 [~dhruv13@84.64.13.237] has joined #shogun | 20:04 | |
pickle27 | sup | 20:10 |
pickle27 | wanna deploy the site now sonney2k | 20:10 |
-!- sonne|osx [~sonne@e177086223.adsl.alicedsl.de] has joined #shogun | 20:13 | |
sonne|osx | pickle27: hey there | 20:15 |
pickle27 | hey | 20:15 |
sonne|osx | pickle27: maybe you want to do the upgrades yourself? | 20:16 |
PirosB3 | Hey guys, I just posted on the ML a new restructured version of the Shogun SaaS spec. If anyone wants to have a look and has some comments, be my guest :) The doc is available here: https://docs.google.com/document/d/1KISZL6lyeR-RvWEycTKqOunr3gwrA3PUnxZ8hiJYfgw/edit#heading=h.k3bcyd74up5z | 20:16 |
pickle27 | Im pretty indifferent, you were able to do them pretty quick before | 20:16 |
pickle27 | its just 2 simple evolutions and then we're done | 20:17 |
sonne|osx | pickle27: well it makes sense if you do the releases of the webstuff in the future yourself anyways - right? | 20:19 |
pickle27 | yeah sure | 20:19 |
pickle27 | what do I need? | 20:19 |
pickle27 | also I lost that ssh key when I re did my home computer | 20:19 |
@HeikoS | sonne|osx: getting food now, back in 40 mins | 20:25 |
sonne|osx | pickle27: okido then please guide me what to do | 20:25 |
-!- kislay [~kislay@93.114.45.13] has quit [Ping timeout: 244 seconds] | 20:25 | |
pickle27 | so deploy the latest code | 20:26 |
pickle27 | then run | 20:26 |
pickle27 | ./manage evolve --execute | 20:26 |
-!- iglesiasg [~iglesiasg@524AE0A7.cm-4-3d.dynamic.ziggo.nl] has joined #shogun | 20:31 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 20:31 | |
@iglesiasg | Goedenavond :) | 20:31 |
-!- Cheparukhin [b064f6fe@gateway/web/freenode/ip.176.100.246.254] has quit [Ping timeout: 245 seconds] | 20:32 | |
-!- Osho_ [0e8b2998@gateway/web/freenode/ip.14.139.41.152] has joined #shogun | 20:34 | |
sonne|osx | ok I did evolve | 20:34 |
pickle27 | everything ran good? | 20:35 |
sonne|osx | well evolutions successfol | 20:35 |
pickle27 | okay sweet! | 20:35 |
sonne|osx | ok restarting servers | 20:35 |
pickle27 | ohhhh | 20:36 |
pickle27 | okay I tried to load something | 20:36 |
pickle27 | and it was all down | 20:36 |
pickle27 | and I was freaking out for a sec | 20:36 |
sonne|osx | pickle27: looks good to me | 20:36 |
sonne|osx | pickle27: some minor thing - on the thumbnail scroll thing at the bottom we have some overlap of the cursor with the image | 20:37 |
pickle27 | subpages aren't loading but I think we just need to merge the next 2 PRs | 20:37 |
sonne|osx | pickle27: ohh ok | 20:38 |
-!- kislay [~kislay@93.114.45.13] has joined #shogun | 20:38 | |
sonne|osx | pickle27: then be quick! | 20:39 |
pickle27 | sonne|osx: can you run a ./manage syncdb | 20:39 |
sonne|osx | pickle27: well I did | 20:40 |
pickle27 | like just now? | 20:40 |
sonne|osx | Creating tables ... | 20:40 |
sonne|osx | Installing custom SQL ... | 20:40 |
sonne|osx | Installing indexes ... | 20:40 |
sonne|osx | Installed 0 object(s) from 0 fixture(s) | 20:40 |
pickle27 | hmm | 20:41 |
-!- Osho_ [0e8b2998@gateway/web/freenode/ip.14.139.41.152] has left #shogun [] | 20:41 | |
pickle27 | cause whats happening is that there is a column in the db that django doesn't understand | 20:41 |
pickle27 | but that column was supoosed to be removed by the evolution | 20:41 |
-!- kislay [~kislay@93.114.45.13] has quit [Read error: Connection reset by peer] | 20:42 | |
-!- kprah [67157e4c@gateway/web/freenode/ip.103.21.126.76] has joined #shogun | 20:43 | |
pickle27 | sonne|osx: what was the output from the evolution? | 20:44 |
pickle27 | can you try and run them again? | 20:44 |
kprah | wiking: hey! | 20:46 |
@sonney2k | pickle27, then it says No evolution required. | 20:47 |
pickle27 | damn | 20:47 |
pickle27 | I mean I can add the line back to the model and it will stop crashing | 20:48 |
pickle27 | but deleting that old foreign key was the whole point | 20:49 |
kprah | \msg wiking Should I unify all the bundle solvers : primal and bmrm to a single cutting plane solver? | 20:49 |
pickle27 | sonney2k: we have 2 options | 20:51 |
pickle27 | we can put the line in the modelback so it knows what to do with that field | 20:51 |
pickle27 | or we drop the column manually | 20:51 |
pickle27 | I don't know if django will like that | 20:51 |
sonne|osx | pickle27: what would I need to run? | 20:54 |
pickle27 | you'd have to open mysql and run | 20:54 |
pickle27 | ALTER TABLE pages_articles DROP COLUMN rootsubpage | 20:54 |
pickle27 | if that fails we'll probably have to restore the db | 20:54 |
sonne|osx | ALTER TABLE pages_articles DROP COLUMN rootsubpage; | 20:56 |
sonne|osx | ERROR 1146 (42S02): Table 'shogun.pages_articles' doesn't exist | 20:56 |
pickle27 | okay I guess the table name wrong | 20:56 |
pickle27 | can you print a schema of the db? | 20:56 |
PirosB3 | need some Django help guys? | 20:57 |
pickle27 | sure feel free to help out | 20:57 |
pickle27 | I ran an evolution to drop an old column | 20:57 |
pickle27 | but its not happy about it | 20:58 |
PirosB3 | what migration app are you using? south? | 20:58 |
-!- drbabayeve [~drbabayev@193.104.110.150] has joined #shogun | 20:58 | |
pickle27 | evolution | 20:58 |
PirosB3 | ash, never used | 20:59 |
pickle27 | I think we need to dump the db reset and restore | 21:00 |
-!- iglesiasg [~iglesiasg@524AE0A7.cm-4-3d.dynamic.ziggo.nl] has quit [Ping timeout: 244 seconds] | 21:01 | |
PirosB3 | can you paste bin the evolution file? | 21:01 |
PirosB3 | and the error? | 21:01 |
PirosB3 | also, is the schema available online somewhere? | 21:01 |
pickle27 | https://github.com/shogun-toolbox/shogun-web/commit/14eb44b62c961da9112d6b4beb61b76076ea8a89 | 21:01 |
pickle27 | the evolution runs fine | 21:01 |
pickle27 | the issue is django is still finding the field somehow | 21:01 |
pickle27 | \and then its not on the model so its throwing an error | 21:01 |
PirosB3 | mm interesting.. | 21:01 |
PirosB3 | did you grep for the field name? | 21:02 |
PirosB3 | but did the migration finish successfully? | 21:02 |
-!- kprah [67157e4c@gateway/web/freenode/ip.103.21.126.76] has left #shogun [] | 21:04 | |
pickle27 | I have not greped for the field name | 21:04 |
pickle27 | yes the evolution ran successfuly | 21:04 |
PirosB3 | interesting, so where is it failing | 21:06 |
PirosB3 | can you paste the trace? | 21:06 |
pickle27 | wait | 21:06 |
pickle27 | I might have found it | 21:06 |
pickle27 | sonney2k: | 21:07 |
pickle27 | found it | 21:07 |
pickle27 | its a stray field | 21:07 |
pickle27 | thanks PirosB3 | 21:07 |
PirosB3 | no worries :) glad you found it ;) | 21:07 |
sonne|osx | pickle27: so now what? | 21:07 |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 21:07 | |
shogun-notifier- | shogun-web: Kevin Hughes :master * 12db238 / pages/admin.py,pages/models.py: https://github.com/shogun-toolbox/shogun-web/commit/12db238950528b6e01b7990642038e6427c7c4de | 21:07 |
shogun-notifier- | shogun-web: removes legacy models | 21:07 |
shogun-notifier- | shogun-web: Kevin Hughes :master * 581ec39 / pages/admin.py,pages/models.py: https://github.com/shogun-toolbox/shogun-web/commit/581ec394389416a02ed75d8e6b1d4b83e5224c80 | 21:07 |
shogun-notifier- | shogun-web: Merge pull request #58 from pickle27/drop_tables_time | 21:07 |
shogun-notifier- | shogun-web: | 21:07 |
shogun-notifier- | shogun-web: removes legacy models | 21:07 |
pickle27 | sonney2k: can you deploy latest code | 21:08 |
@HeikoS | heya! | 21:08 |
pickle27 | it should be fine | 21:08 |
pickle27 | after that | 21:08 |
lisitsyn | hey guys I have something to say about plugins :D | 21:08 |
lisitsyn | who cares? ;) | 21:08 |
pickle27 | sonney2k: after deploy run ./manage syncdb | 21:09 |
pickle27 | it should drop the old tables | 21:09 |
@HeikoS | lisitsyn: I dont | 21:10 |
@HeikoS | hahaha ;) | 21:10 |
lisitsyn | haha I knew | 21:10 |
@HeikoS | lisitsyn: no please shoot I want to have that | 21:11 |
@HeikoS | lisitsyn: loosing too much lifetime waiting for shogun to compile | 21:11 |
-!- kprah [67157d4c@gateway/web/freenode/ip.103.21.125.76] has joined #shogun | 21:11 | |
lisitsyn | okay I actually started something as I have absolutely no clue | 21:11 |
lisitsyn | https://github.com/lisitsyn/plugins-playground | 21:11 |
lisitsyn | it is quite easy to load library and stuff | 21:11 |
lisitsyn | current version just loads some .so dynamically and calls method | 21:12 |
lisitsyn | so the thing is that how do we go next | 21:12 |
lisitsyn | basically what I see is | 21:12 |
lisitsyn | we have to have some base classes | 21:12 |
lisitsyn | as classifier multiclass classifier etc | 21:12 |
@HeikoS | lisitsyn: oh wow | 21:12 |
@HeikoS | cool that already works | 21:12 |
lisitsyn | but we'd have to stick to some get("parametername",value); | 21:13 |
lisitsyn | otherwise we can't get getter | 21:13 |
lisitsyn | :D | 21:13 |
lisitsyn | let me explain | 21:13 |
@HeikoS | lisitsyn: is that now parameter framework or plugins? | 21:13 |
@HeikoS | or both at once? | 21:13 |
lisitsyn | HeikoS: they are pretty much linked together it seems | 21:13 |
lisitsyn | so what I mean is | 21:13 |
lisitsyn | imagine we have super svm | 21:13 |
@HeikoS | lisitsyn: mmh so those getters are like ultra slow right? | 21:13 |
lisitsyn | and it has parameter heiko | 21:13 |
@HeikoS | lisitsyn: what to do with iterating over matrices? | 21:14 |
@HeikoS | and stuff like that? | 21:14 |
@HeikoS | or do we only do that for CSGObject like getters? | 21:14 |
@HeikoS | or how? | 21:14 |
lisitsyn | that's the question | 21:14 |
lisitsyn | so our super svm has parameter heiko = 3 | 21:14 |
@HeikoS | access should be O(1) not O(1000) | 21:14 |
-!- iglesiasg [~iglesiasg@524AE0A7.cm-4-3d.dynamic.ziggo.nl] has joined #shogun | 21:14 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 21:14 | |
lisitsyn | haha O(1000) | 21:14 |
@HeikoS | iglesiasg: greetings! | 21:14 |
@iglesiasg | HeikoS, hello hello | 21:15 |
lisitsyn | HeikoS: that's the thing | 21:15 |
@HeikoS | lisitsyn: mmh but do we actually call getters in loops? | 21:15 |
lisitsyn | so well it is the question whether we call get/set pretty often | 21:15 |
@HeikoS | lisitsyn: one can always call it before right? | 21:15 |
lisitsyn | well yeah | 21:15 |
@HeikoS | lisitsyn: and iterating over matrices etc still works directly right? | 21:15 |
lisitsyn | yeah this is not about matrices | 21:16 |
lisitsyn | matrices are not modular at all they are still in core | 21:16 |
lisitsyn | HeikoS: actually | 21:16 |
lisitsyn | what we'd need is | 21:16 |
lisitsyn | to replace all getters in classes | 21:16 |
lisitsyn | which are plugins | 21:16 |
@HeikoS | lisitsyn: yeah I see | 21:16 |
lisitsyn | see what I mean? | 21:16 |
lisitsyn | anything else is ok | 21:16 |
@HeikoS | lisitsyn: so anyone who attempts this will commit suicide right? | 21:16 |
lisitsyn | why? | 21:17 |
shogun-notifier- | shogun-web: Kevin Hughes :master * 9e8b87f / pages/models.py: https://github.com/shogun-toolbox/shogun-web/commit/9e8b87f55eba7dc74dfd8705f5b22b9706b95082 | 21:17 |
shogun-notifier- | shogun-web: remove rootsubpage | 21:17 |
lisitsyn | the problem is that once we loaded a plugin | 21:17 |
@HeikoS | lisitsyn: actually, I think computationally it should be fine, why would one call getters in a loop? | 21:17 |
lisitsyn | and spawned its say classifier | 21:17 |
pickle27 | sonney2k: would help if I actually fixed it lol | 21:17 |
lisitsyn | HeikoS: I can make it quite fast | 21:17 |
@HeikoS | lisitsyn: how? hashs? | 21:17 |
lisitsyn | yeah | 21:17 |
@HeikoS | lisitsyn: man that would be so cool | 21:17 |
lisitsyn | one downside that I see | 21:18 |
lisitsyn | we'd get some string things | 21:18 |
lisitsyn | libsvm = shogun.require('LibSVM') | 21:18 |
lisitsyn | that's like handle for library | 21:18 |
lisitsyn | or plugin | 21:18 |
lisitsyn | next what we do is | 21:19 |
lisitsyn | svm = libsvm.spawnClassifier() | 21:19 |
lisitsyn | or sth like that | 21:19 |
-!- kprah_ [67157f4c@gateway/web/freenode/ip.103.21.127.76] has joined #shogun | 21:19 | |
lisitsyn | to do something with hashing | 21:19 |
lisitsyn | we can do this thing: | 21:19 |
lisitsyn | svm_C_parameter = shogun.parameter('C') | 21:20 |
-!- kprah [67157d4c@gateway/web/freenode/ip.103.21.125.76] has quit [Ping timeout: 245 seconds] | 21:20 | |
lisitsyn | svm.get(svm_C_parameter); | 21:20 |
lisitsyn | oh sorry | 21:20 |
lisitsyn | it should be shogun.floatParameter('C') | 21:20 |
lisitsyn | like that | 21:20 |
lisitsyn | this object would have some hash or index | 21:20 |
lisitsyn | so get is still fast | 21:20 |
lisitsyn | HeikoS: are you following this sht? :D | 21:20 |
@HeikoS | lisitsyn: yeah I am sorry | 21:21 |
lisitsyn | no I just mean should I stop at this point | 21:21 |
lisitsyn | or continue some explanations hehe | 21:21 |
@HeikoS | lisitsyn: could you set up a document where you write example (pseudo) code how things are used? | 21:21 |
@HeikoS | lisitsyn: its hard to loose overview | 21:21 |
lisitsyn | yeah | 21:21 |
@HeikoS | this spawn thing, I did not really get it | 21:21 |
lisitsyn | it comes directly from my mind yet | 21:21 |
lisitsyn | HeikoS: well ok that needs explanation here I guess | 21:22 |
lisitsyn | HeikoS: once we loaded a library | 21:22 |
lisitsyn | we'd have to keep it as an object in our scope | 21:22 |
lisitsyn | however that's okay if you just recall it next with | 21:22 |
lisitsyn | shogun.module('LibSVM') | 21:22 |
lisitsyn | HeikoS: the point is that if we have plugins we don't have qualified names in the scope | 21:23 |
lisitsyn | we don't have shogun.classifier.LibSVM anymore | 21:23 |
@HeikoS | lisitsyn: cant we enforce those? | 21:23 |
-!- kprah_ [67157f4c@gateway/web/freenode/ip.103.21.127.76] has quit [Ping timeout: 245 seconds] | 21:23 | |
@HeikoS | in a tree like manner? | 21:23 |
lisitsyn | HeikoS: what do you mean? | 21:24 |
lisitsyn | it would be possible for python | 21:24 |
lisitsyn | but java etc | 21:24 |
lisitsyn | I don't like altering behaviour in various languages | 21:24 |
@HeikoS | yeah I agree | 21:24 |
@HeikoS | lisitsyn: ok | 21:24 |
@HeikoS | lisitsyn: actually I think you just should do this, discuss with somebody who wants to spend lots of brainpower on it | 21:25 |
@HeikoS | lisitsyn: unlikely to get much feedback from the other core people ;) | 21:25 |
lisitsyn | HeikoS: I'll dump my mind laterz | 21:25 |
lisitsyn | to some issue | 21:25 |
@HeikoS | lisitsyn: maybe a list of pros/cons would be nice too | 21:25 |
@HeikoS | lisitsyn: I currently dont really have an idea how this works in principle | 21:25 |
lisitsyn | yeah but this looks like a complete solution in my mind | 21:25 |
lisitsyn | :D | 21:25 |
@HeikoS | lisitsyn: well I pretty much trust you :D | 21:26 |
-!- drbabayeve [~drbabayev@193.104.110.150] has quit [Remote host closed the connection] | 21:27 | |
@HeikoS | lisitsyn: anybody yet interested in that? | 21:28 |
lisitsyn | HeikoS: gsoc? | 21:28 |
@HeikoS | lisitsyn: maybe you should make it a bit more attractive in the description. | 21:28 |
lisitsyn | no that's too hardcore | 21:28 |
lisitsyn | :D | 21:28 |
lisitsyn | yeah my bad may be | 21:28 |
@HeikoS | lisitsyn: Its one of the most interesting projects | 21:28 |
@HeikoS | lisitsyn: and its not really ML, so that is interesting | 21:28 |
@HeikoS | but also to get to know shogun | 21:29 |
@HeikoS | unusual coding tasks | 21:29 |
@HeikoS | etc | 21:29 |
lisitsyn | well I know a student that I can guide directly :D | 21:29 |
@HeikoS | lisitsyn: ok? | 21:30 |
@HeikoS | tell me more | 21:30 |
lisitsyn | HeikoS: just kidding | 21:30 |
lisitsyn | :D | 21:31 |
@HeikoS | we should soon get applications btw :) | 21:31 |
@HeikoS | lisitsyn: yourself? | 21:31 |
@HeikoS | haha | 21:31 |
lisitsyn | HeikoS: quite near | 21:31 |
@HeikoS | noooo way man, too precious | 21:31 |
@HeikoS | lisitsyn: ah your girl? ;) | 21:31 |
pickle27 | sonne|osx: have you deployed again yet? | 21:31 |
lisitsyn | haha yeah but this could be considered unfair | 21:31 |
-!- PirosB3 [~pirosb3@162.219.6.191] has quit [Quit: PirosB3] | 21:32 | |
@HeikoS | hehe is she a c++ guru? | 21:32 |
lisitsyn | HeikoS: no no | 21:32 |
@HeikoS | lisitsyn: btw will you come to the summer workshop this year? | 21:33 |
lisitsyn | HeikoS: yeah if I can afford it | 21:33 |
@HeikoS | nice | 21:34 |
@HeikoS | lisitsyn: how do you feel about our gsoc status currently? | 21:35 |
lisitsyn | HeikoS: that's okay | 21:35 |
lisitsyn | we have some strong students | 21:35 |
-!- abinash_panda [0126170c@gateway/web/freenode/ip.1.38.23.12] has quit [Ping timeout: 245 seconds] | 21:36 | |
@HeikoS | lisitsyn: yeah I agree | 21:36 |
@HeikoS | looking good | 21:36 |
@HeikoS | and the niveau is insane compared to other years, at least I have the feeling | 21:37 |
shogun-notifier- | shogun-web: Kevin Hughes :master * 2bc3934 / pages/ (8 files): https://github.com/shogun-toolbox/shogun-web/commit/2bc3934cb8d02b36d1133dee63e2a6d6ad52ca8b | 21:37 |
shogun-notifier- | shogun-web: roll up the db changes etc | 21:37 |
shogun-notifier- | shogun-web: Kevin Hughes :master * 86622f7 / pages/ (8 files): https://github.com/shogun-toolbox/shogun-web/commit/86622f7aa938ef07ccdeeafd844e8df67cb3eaf4 | 21:37 |
shogun-notifier- | shogun-web: Merge pull request #59 from pickle27/rollup | 21:37 |
shogun-notifier- | shogun-web: | 21:37 |
shogun-notifier- | shogun-web: Rollup | 21:37 |
-!- HeikoS [~heiko@p4FED46C2.dip0.t-ipconnect.de] has quit [Ping timeout: 240 seconds] | 22:03 | |
-!- HeikoS [~heiko@ra-vpn92.ra-vpn.ucl.ac.uk] has joined #shogun | 22:18 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 22:18 | |
lisitsyn | HeikoS: I noticed you have been online from some german address ;) | 22:26 |
lisitsyn | and now it is uk again | 22:26 |
lisitsyn | :D | 22:26 |
shogun-notifier- | shogun-web: Kevin Hughes :master * 815047b / pages/models.py: https://github.com/shogun-toolbox/shogun-web/commit/815047bdf604dfbc3a9046ad74d3b05f4885913d | 22:31 |
shogun-notifier- | shogun-web: add fake rootpage integer field | 22:31 |
@sonney2k | lisitsyn, heiko the tunnel guy | 22:39 |
lisitsyn | haha | 22:39 |
@sonney2k | yay python shogun 3.2.0 uploaded https://buildd.debian.org/status/package.php?p=python-shogun | 22:40 |
@sonney2k | lets see what fails :) | 22:40 |
-!- [paladin] [73f91219@gateway/web/freenode/ip.115.249.18.25] has quit [Quit: Page closed] | 22:47 | |
-!- pickle27 [d8bfe7da@gateway/web/freenode/ip.216.191.231.218] has quit [Quit: Page closed] | 22:48 | |
@sonney2k | puhh website is back up | 22:52 |
@sonney2k | what a stress... | 22:52 |
-!- iglesiasg [~iglesiasg@524AE0A7.cm-4-3d.dynamic.ziggo.nl] has quit [Quit: Leaving] | 22:55 | |
@HeikoS | sonney2k, lisitsyn yeah tunnel :) | 22:56 |
@HeikoS | working on cluister computers | 22:56 |
-!- HeikoS [~heiko@ra-vpn92.ra-vpn.ucl.ac.uk] has quit [Read error: Connection reset by peer] | 23:00 | |
-!- HeikoS [~heiko@p4FED46C2.dip0.t-ipconnect.de] has joined #shogun | 23:20 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 23:20 | |
sonne|osx | besser82: ok maybe it is indeed a bug in numpy that we ahve the failure on powerpc https://buildd.debian.org/status/package.php?p=python-shogun | 23:35 |
-!- sonne|osx [~sonne@e177086223.adsl.alicedsl.de] has quit [Quit: sonne|osx] | 23:44 | |
-!- HeikoS [~heiko@p4FED46C2.dip0.t-ipconnect.de] has quit [Quit: Leaving.] | 23:55 | |
-!- HeikoS [~heiko@p4FED46C2.dip0.t-ipconnect.de] has joined #shogun | 23:57 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 23:57 | |
--- Log closed Tue Mar 11 00:00:33 2014 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!